Deepseek wants to take on Claude Code and OpenAI's Codex with "Deepseek Code"

2026-05-20 · Source: The Decoder · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering, Emerging Technologies & Innovation · Depth: Intermediate, quick

Summary

Deepseek, a Chinese AI company, is establishing a new "Harness" team in Beijing to develop "Deepseek Code," a direct competitor to Anthropic's Claude Code and OpenAI's Codex. Announced on May 20, 2026, by Deli Chen from Deepseek, this project focuses on building a comprehensive AI agent where the core model is augmented by a "Harness" layer for tool use, planning, and memory. The company is actively hiring a product manager and a developer, seeking candidates proficient with existing tools like Claude Code, Cursor, and GitHub Copilot, and experienced in agent loops, multi-agent systems, and context engineering. This initiative positions Deepseek Code at the intersection of advanced AI research and practical product development.

Key takeaway

For AI Engineers or Product Managers evaluating the competitive landscape of code generation tools, Deepseek's "Deepseek Code" signals a significant new entrant focusing on comprehensive AI agents. You should monitor its development, particularly the "Harness" approach to tool use and planning, as it could influence future agent architectures. Consider how your current projects might benefit from or need to adapt to more integrated, multi-agent coding assistants.

Key insights

Deepseek is developing "Deepseek Code" as an AI agent by combining a core model with a "Harness" for advanced capabilities.

Principles

• AI agents integrate a core model with a "Harness" layer.
• "Harness" encompasses tool use, planning, and memory.
• Agent development requires expertise in multi-agent systems.

Method

Building an AI agent involves integrating a core model with a "Harness" layer for tool use, planning, and memory capabilities, guided by a product roadmap and community feedback.

In practice

• Deepseek Code aims to compete with Claude Code, OpenAI's Codex, and Cursor.
• Roles require experience with agent loops and context engineering.

Topics

• AI Agents
• Code Generation
• Deepseek Code
• Claude Code
• OpenAI Codex
• Context Engineering
• Multi-agent Systems

Best for: AI Product Manager, AI Engineer, Director of AI/ML

Related on AIssential

• Claude Code and What Comes Next — New AI coding tools combine enhanced autonomy with agentic harnesses to perform complex, sustained tasks.
• DeepSeek V4 + Claude Code = BEST AI Coder! — Combine DeepSeek V4 with Claude Code for a cost-effective, hybrid AI coding workflow.
• 👨‍🔧 OpenAI launches Codex app to gain ground in AI coding race — AI coding tools enhance speed but may hinder skill development if over-relied upon for core tasks.
• [D] We reimplemented Claude Code entirely in Python — open source, works with local models — A Python-based, open-source Claude Code agent reimplementation enables local execution and extensibility.
• Fragments: April 21 — AI's rapid advancement necessitates revisiting foundational software principles and developing robust security for "permission hungry" agents.
• [AINews] Every Lab serious enough about Developers has bought their own Devtools — AI development is converging on agentic coding, multi-agent systems, and enterprise-grade tooling, driven by strategic acquisitions and product innovations.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by The Decoder.